Pulsars… Neutron Star Astrophysics Pulsar Sounds

Pulsars I.
Pulsars I.
The Why and How of Searching for Exotic Pulsars
The Why and How of Searching for Exotic Pulsars
Jim Cordes, Cornell University
Jim Cordes, Cornell University
• Why would you want to know about pulsars and
why would you like to discover more?
• Science, the big questions
• How do I find new pulsars?
• How do I estimate pulsar distances?
• What can I learn about …
•
•
•
•
•
• The forefront of neutron star science
•
•
•
•
Extreme states of matter
Gravitational laboratories
Probing core collapse supernovae
Galactic structure
• Issues in pulsar survey optimization
•
•
•
•
the inside of a NS?
the magnetosphere of a NS?
the orbit of a binary pulsar?
the space velocity of a pulsar?
the ISM along the path to a pulsar?
Dedispersion
Periodicity searches
Single pulse searches
Simulations of pulsar surveys
• ALFA: A massive pulsar survey at Arecibo
• SKA: toward a full Galactic census of pulsars
Pulsars…
Neutron Star Astrophysics
• Embody physics of the EXTREME
– surface speed ~0.1c
– 10x nuclear density in center
– some have B > Bq = 4.4 x 1013 G
– Voltage drops ~ 1012 volts
– FEM = 109Fg = 109 x 1011FgEarth
– Tsurf ~ 106 K
• Relativistic plasma physics in action (γ ~ 106)
• Probes of turbulent and magnetized ISM
• Precision tools, e.g.
- Period of B1937+21 (the fastest millisecond pulsar)
P = 0.0015578064924327±0.0000000000000004 s
Orbital eccentricity of J1012+5307: e<0.0000008
• Laboratories for extreme states of matter and as clocks for
probing space-time and Galactic material
Surface quantities:
B ≈ 1012 Gauss
gNS ≈ 1011 g⊕
FEM ≈ 109 gNS mp
Φ ≈ 1012 volts
Pulsar Sounds
Radio signals demodulated into audio signals
P
(ms)
f=1/P
(Hz)
B0329+54
714
1.4
B0950+08
253
3.9
B0833-45
(Vela)
89
11.2
B0531+21
(Crab)
33
30.2
5.7
174
1.56
642
Pulsar
J0437-4715
B1937+21
sound file
J0437J0437-4715
profile
1
Pulsar Populations:
diagram
• Canonical
Pulsar Populations:
• Millisecond pulsars (MSPs)
• P ~ 1.5 – 20ms
• B ~ 108 – 109 ms
• High field
• P~5–8s
1013
G
• Braking index n:
•
• Millisecond pulsars (MSPs)
• P ~ 1.5 – 20ms
• B ~ 108 – 109 ms
• High field
• P~5–8s
• B ~ few x 1013 G
• Braking index n:
•
• n=3 magnetic dipole radiation
log Period derivative (s s-1)
• P~ 20ms – 5s
• B ~ 1012±1 G
log Period derivative (s s-1)
• P~ 20ms – 5s
• B ~ 1012±1 G
• B ~ few x
diagram
• Canonical
• n=3 magnetic dipole radiation
• Death line
• Death line
• Strong selection effects
• Strong selection effects
Period (sec)
Period (sec)
Manifestations of NS:
Spindown
• Rotation driven:
• “radio” pulsars (radio → γ rays)
• magnetic torque
• γγ → e+ e- + plasma instability ⇒ coherent radio
• Accretion driven:
• X-rays
• LMXB, HMXB
• Magnetic driven: Crustquakes?
• Magnetars (AXPs, SGRs)
• Spindown … but
• Gravitational catastrophes?
• Gamma-ray bursts, G.wave sources, hypernovae?
Quarks to Cosmos: Relevant Questions
Spinup
•
Did Einstein have the last word on
Gravity?
•
How do cosmic accelerators work
and what are they accelerating?
•
What are the new states of matter at
exceedingly high density and
temperature?
•
Is a new theory of light and matter
needed at the highest energies?
2
First Double Pulsar: J0737-3939
Lyne et al.(2004)
• Pb=2.4 hrs, dω/dt=17 deg/yr
• MA=1.337(5)M, MB=1.250(5)M
Now to 0.1%
Double Neutron Star Binary:
J0737-0939A,B
Testing GR:
s obs
= 1.000 ± 0.002
s exp
Kramer et al.(2004)
What are the Big Questions?
• Formation and Evolution:
• What determines if a NS is born as a magnetar vs a
canonical pulsar?
• How fast do NS spin at birth?
• How fast can recycled pulsars spin?
• What is the role of instabilities and gravitational radiation in
determining the spin state?
• How do momentum thrusts during core collapse affect the
resulting spin state and translational motion of the NS?
• What processes determine the high space velocities of NS?
»
»
»
»
Neutrino emission
Matter rocket effects
Electromagnetic rocket effect (Harrison-Tademaru)
Gravitational-wave rocket effect
• Are orbital spiral-in events at all related to high-energy
bursts? (GRBs? Other transients?)
Bow Shocks
Bow Shocks
MSPs
Palomar
H image
Guitar Nebula:
Low V
• Ordinary pulsar
High
Edot
• P = 0.68 s
B1957+20 (Kulkarni & Hester; Gaensler et al.
J0437-47
(Fruchter et al.)
J2124-3358
Gaensler et al
Mouse
Duck
RXJ1856
B0740-28
Canonical pulsars
High V, low to high Edot
J0617
• B = 2.6 x 1012 G
• = 1.1 Myr
•E=I
1033.1 erg s-1
• D 1.9 kpc (from DM)
• 1600 km s-1 at nominal distance
• Will escape the Milky Way
HST
WFPC2
H
1994
Radius of curvature of bowshock
nose increased from 1994 to
2001, corresponding to a 33%
decrease in ISM density
The pulsar is emerging from a
region of enhanced density
2001
Chatterjee & Cordes 2004
3
What are the Big Questions?
• NS Structure:
•
•
•
•
Are neutron stars really neutron stars?
What comprises the core of a NS?
What is the mass distribution of NS?
In what regions are the neutrons in a superfluid
state?
• How large are interior magnetic fields?
• Magnetosphere and Emission Physics:
• What QED processes are relevant for
electromagnetic emissions?
Forefronts in NS Science
• Understanding NS populations and their
physical differences
•
•
•
•
Radio pulsars and their progenitors
Magnetars
Radio quiet/Gamma-ray loud objects
Branching ratios in supernovae
• The physics of NS runaway velocities
• Are “neutron stars” neutron stars?
Fulfilling the Promise of NS Physics
and Astrophysics
•
•
•
•
What are the Big Questions?
• NS as Laboratories:
• Can departures from General Relativity be identified in the orbits
of compact binary pulsars?
• Does the Strong-Equivalence Principle hold to high precision in
pulsars with WD or BH companions?
• NS as Gravitational Wave Detectors:
• Use pulsars to detect long-period gravitational waves
» Early universe
» Mergers of supermassive black holes
» Topological defects (cosmic strings)
• Pulsars as Probes of Galactic Structure
• What kind of spiral structure does the Galaxy have?
• What is the nature of interstellar turbulence?
Forefronts in NS Science
• Finding compact relativistic binary pulsars for
use as laboratories
• Gravity
• Relativistic plasma physics in strong B
• Finding spin-stable MSPs for use as
gravitational wave detectors (λ ~ light years)
• h ~ σTOA T-1 (T = data span length)
• Complete surveys of the transient radio sky
• pulsars as prototype coherent radio emission
Currently about 1700 pulsars known
Galactic birth rate ~ 1/100 yr × 10 Myr lifetime for canonical pulsars
⇒105 active pulsars × 20% beaming fraction
⇒ 2×104 detectable pulsars + 10% more MSPs + NS-NS, NS-BH etc.
Find more pulsars
Time them with maximal precision
Phase-resolved polarimetry
VLBI them to get high astrometric precision
Step 1: conduct surveys that optimize the
detection of faint, pulsed emission that is
dispersed and that may or may not be periodic.
4
Arecibo + SKA Surveys
Pulsar Search Domains
Region/Direction
A Single Dispersed Pulse from the Crab Pulsar
Kind of Pulsar
Telescope
Arecibo,
Effelsburg, GBT,
Jodrell, Parkes,
WSRT
Galactic Plane
Young pulsars
(< 1 Myr)
Galactic Center
Young, recycled,
GBT, SKA
binary, circum-SgrA*
Moderate Galactic
latitudes
MSPs, binary,
runaway
Arecibo, GBT,
Parkes
Globular clusters
MSPs, binary
Arecibo, GBT,
Parkes
Local Group
Galaxies
Young (probably)
Giant pulses
Arecibo, GBT,
SKA
Refractive indices for cold, magnetized plasma
2
3
nl,r ~ 1 - νp2 / 2ν 2 −
+ νp νB 2ν
Arecibo WAPP
ν >> νp ~ 2 kHz
S ~ 160 x Crab Nebula
~ 200 kJy
ν >> νB ~ 3 Hz
Group velocity ⇒ group delay = ∆ (time of arrival)
t = t DM ± t RM
Detectable to ~ 1.5 Mpc
with Arecibo
birefringence
t DM = 4.15 ms DM ν
-2
t RM = 0.18 ns RM ν
-3
Dispersion Measure DM = ∫ ds ne pc cm-3
Rotation Measure
Refractive indices for cold, magnetized plasma
m
n >> np ~ 2 kHz
np2
n B 2n
n >> n B ~ 3 Hz
Group velocity ⇒ group delay = ∆ (time of arrival)
t = t DM ± t RM
t DM = 4.15 ms DM ν
t RM = 0.18 ns RM ν
DM = ∫ ds ne
-2
-3
pc cm-3
RM = 0.81 ∫ ds ne B rad m-2
Basic data unit = a dynamic spectrum
3
106 – 108 samples x 64 µs
Fast-dump
spectrometers:
Frequency
/ 2n
2
64 to 1024 channels
nl,r ~ 1 -
np2
RM = 0.81 ∫ ds ne B rad m-2
P
time
•
Analog filter banks
•
Correlators
•
FFT (hardware)
•
FFT (software)
•
Polyphase filter bank
E.g. WAPP, GBT
correlator + spigot
card, new PALFA
correlator
5
Frequency
Frequency
P
P
Time
Time
Interstellar Scintillation
RFI
New Pulsars
Periodicity
Search
Known Pulsars
Single-pulse search
(matched filtering)
Arrival time
Monitoring
Polarization
Analysis
Scintillation
Studies
(FFT)
Frequency
Astrophysical effects are typically buried in noise and RFI
Issues In Pulsar Survey Optimization
P
• Broad luminosity function for pulsars
• Beam luminosity
• Geometric beaming
Time
• Pulse sharpness
• Intrinsic pulse width W
• Smearing from propagation effects
New Pulsars
Periodicity
Search
Single-pulse search
(matched filtering)
» Dispersion across finite bandwidth (correctable)
» Multipath propagation (scattering in the ISM)
Known Pulsars
Arrival time
Monitoring
Polarization
Analysis
Scintillation
Studies
(FFT)
• Smearing from orbital acceleration
• Intermittency of the pulsar signal
• Nulling, giant pulses, precession, eclipsing
• Interstellar scintillation
Dedispersion
Issues In Pulsar Survey Optimization
• Combine the signal over time and frequency while
maximizing S/N through matched filtering:
• Dedispersion: sum over frequency while removing the dispersive
time delays
• Single pulses: match the shape and width of the pulse
• Periodic pulses: match the period as well as the pulse shape and
width
• Orbital motion: match the change in pulse arrival times related to
the changing Doppler effect
⇒ Single-pulse searches:
Search vs. (DM, W)
⇒ Periodicity searches:
Two methods:
Coherent:
• operates on the voltage proportional to the electric
field accepted by the antenna, feed and receiver
• computationally intensive because it requires
sampling at the rate of the total bandwidth
• “exact”
Post-detection:
• operates on intensity = |voltage|2
• computationally less demanding
• an approximation
Search vs. (DM, W, P, [orbital parameters])
6
Basic data unit = a dynamic spectrum
Dispersed Pulse
Coherently dedispersed pulse
Fast-dump
spectrometers:
Frequency
64 to 1024 channels
106 – 108 samples x 64 µs
P
time
•
Analog filter banks
•
Correlators
•
FFT (hardware)
•
FFT (software)
•
Polyphase filter bank
∆ t = 8.3 µ s DM ν-3 ∆ν
E.g. WAPP, GBT
correlator + spigot
card, new PALFA
correlator
Coherent Dedispersion
pioneered by Tim Hankins (1971)
Dispersion delays in the time domain represent a phase
perturbation of the electric field in the Fourier domain:
Coherent Dedispersion
pioneered by Tim Hankins (1971)
Coherent dedispersion works by explicit deconvolution:
Coherent dedispersion involves multiplication of Fourier
amplitudes by the inverse function,
Comments and Caveats:
For the non-uniform ISM, we have
• Software implementation with FFTs to accomplish deconvolution (Hankins 1971)
• Hardware implementations: real-time FIR filters (e.g. Backer et al. 1990s-present)
which is known to high precision for known pulsars.
The algorithm consists of
• Resulting time resolution: 1 / (total bandwidth)
Application requires very fast sampling to achieve usable
bandwidths.
• Actual time resolution often determined by interstellar scattering (multipath)
• Requires sampling at Nyquist rate of 2 samples × bandwidth
⇒Computationally demanding
Micropulses coherently dedispersed
• Most useful for low-DM pulsars and/or high-frequency observations
Nanostructure in Crab pulsar giant pulses
(Hankins1971)
7
Postdetection Dedispersion:
Crab 22-ns resolution
Sum intensity over frequency after correcting for dispersion delay
Interstellar Scattering
Effects
Interstellar scattering from electron density variations
• Pulsar velocities >> ISM, observer velocities
• Scattering is strong for frequencies < 5 GHz
•
•
•
•
•
•
Angular broadening (seeing)
Pulse broadening
Diffractive interstellar scintillations (DISS)
Refractive interstellar scintillations (RISS)
TOA fluctuations (multiple effects)
Superresolution phenomena:
stars twinkle, planets don’t
⇒ pulsars show DISS, AGNs don’t
• Electron density irregularities exist on scales from ~
100’s km to Galactic scales
Pulse broadening from interstellar scattering:
Arecibo WAPP data, Bhat et al 2004
8
Frequency
Scattering limited
Dedispersion at a single known
DM
Frequency
Choose ∆ t ⇒ maximum DM
time
Dispersion limited
time
Σν
I(t)
time
Single Pulse Studies &
Searches
DM
Frequency
Dedispersion over a set of DMs
time
Arecibo WAPP
time
Giant pulse from the
Crab pulsar
Nano-giant pulses (Hankins et al. 2003)
S ~ 160 x Crab Nebula
~ 200 kJy
Arecibo
Detectable to ~ 1.5 Mpc
with Arecibo
2-ns giant pulses from
the Crab: (Hankins et
al. 2003)
Giant Pulses seen
from B0540-69 in
LMC (Johnston &
Romani 2003)
5 GHz
0.5 GHz bw
coherent
dedispersion
9
Giant pulses from M33
Single pulse searches
Arecibo observations
(Mclaughlin & Cordes 2003)
A pulsar found
through its singlesinglepulse emission, not its
periodicity (c.f. Crab
giant pulses).
Pulsar Periodicity
Searches
Algorithm: matched
filtering in the DMDM-t
plane.
ALFA’s 7 beams
provide powerful
discrimination
between celestial and
RFI transients
Example Periodicity Search
Algorithm
DM
Frequency
Pulsar Periodicity Search
time
time
FFT each DM’s
time series
|FFT(f)|
1/P
2/P
3/P
• • •
10
Harmonic Sum
The FFT of periodic pulses is a series of spikes (harmonics) separated by 1/P.
To improve S/N, sum harmonics. This procedure is an approximation to true matched filtering,
which would give optimal S/N.
Sum how many harmonics?
The answer depends on the pulse “duty cycle” = (pulse width / P) (unknown a priori)
⇒ need to use trial values of Nh:
Sum over harmonics
Maximize h() with respect to N h to identify candidate pulsars.
Noise and RFI conspire to yield spurious candidates.
∴ Need a high threshold. How high?
Minimum detectable flux density for a single harmonic:
Minimum detectable flux density for harmonic sum:
Example Time Series and Power Spectrum for a recent PALFA discovery
Example Time Series and Power Spectrum for a recent PALFA discovery
(follow-up data set shown)
(follow-up data set shown)
DM = 0 pc cm-3
Time Series
DM = 217 pc cm-3
Where is the pulsar?
DM = 0 pc cm-3
Time Series
DM = 217 pc cm-3
Here is the pulsar
Pulse
shape
Effects that
broaden
pulses
reduce the
harmonic
sum, which is
bad
FFT
Harmonic
sum
Survey Selection Against Binaries
NS-NS binary
Pulse shape
Phase perturbation
FFT harmonics
Harmonic Sum
11
Dealing With Orbital Motion
Orbital acceleration yields a time-dependent
period, potentially destroying the power of the
straightforward FFT + HS.
How Far Can We Look?
Dmax = D (S / Smin1 )1/2 Nh1/4
Smin1 = single harmonic threshold = m Ssys /(∆ν
∆ν T)1/2
m = no. of sigma ~ 10
• Long-period binaries: T = data span length << Porb
• Do nothing different
Nh = no. of harmonics that maximize
harmonic sum
• Intermediate-period orbits: T < 0.1 Porb
• Acceleration search: compensate the time domain or match filter in
the frequency domain according to an acceleration parameter
• Adds another search parameter: DM, P, W, a
• Very short period orbits: T > Porb (potentially >> Porb)
Nh → 0 for heavily broadened pulses (scattering)
Regimes:
Luminosity limited
Dmax ∝ Smin1 -1/2
DM/SM limited
Dmax ∝ Smin1 -x , x<1/2
• Do conventional FFT but search for orbital sidebands
Dmax vs. Flux Density Threshold
Implications:
• Optimal integration time:
go no deeper than the luminosity limited regime
Scattering
limited
• Fast-dump spectrometers:
need enough channels so that search is not DM limited
Dispersion
limited
Luminosity
limited
• Better to cover more solid angle than to
integrate longer on a given direction
(as long as all solid angles contain pulsars)
AO at S,L,P bands
Add slides showing sensitivity
curves for Arecibo
12
Survey Selection Against Binaries
NS-NS binary
Pulse shape
Hardware for Pulsar Science
Predetection Samplers and Analyzers:
• ASP, GASP (Arecibo & Green Bank)
» Real-time dedispersion and folding
Phase perturbation
• New Mexico Tech burst sampler
» Off-line dedispersion
• Generic baseband samplers (c.f. radar samplers)
FFT harmonics
Postdetection Samplers:
• WAPP (Arecibo), SPIGOT (GBT) (correlators)
» Searching and timing machines
Harmonic Sum
• New PALFA spectrometer (polyphase filter bank)
» Primarily a search machine
Software for Pulsar Searching
•
•
•
•
•
•
Many proprietary packages
Sigproc/Seek package
PRESTO
Cornell Code
Berkeley Code
PALFA: the PALFA Consortium is testing
and consolidating codes to produce a new
“standard” pulsar search package
The power of ALFA:
I(ν
ν , t, θ j) j=1,7
Massive ALFA Pulsar Surveys
103 new pulsars
– Reach edge of Galactic population for much of pulsar luminosity
function
– High sensitivity to millisecond pulsars
– Dmax = 2 to 3 times greater than for Parkes MB
Sensitivity to transient sources
Commensal SETI Search (Wertheimer UCB)
Data management:
– Keep all raw data (~ 1 Petabyte after 5 years) at the Cornell Theory
Center (CISE grant: $1.8M)
– Database of raw data, data products, end products
– Web based tools for Linux-Windows interface (mysql ↔ ServerSql)
– VO linkage (in future)
13
Example of the SKA as a
Pulsar-Search Machine
ALFA pulsar surveys will be the deepest
surveys of the Galaxy until the SKA is built:
~104 pulsar detections with the SKA
(assuming all-sky capability)
• rare NS-NS, NS-BH binaries for
probing strong-field gravity
• millisecond pulsars < 1.5 ms
• MSPs suitable for gravitational wave
detection
• Galactic tomography of electron
density and magnetic field
• Spiral-arm definition
Blue: known pulsars
(prior to Parkes MB)
Red: Parkes MB
Green: PALFA
simulated pulsars
Blue points: SKA simulation
Black points: known pulsars
Pulsar Distances
SKA: What is It?
•
•
•
•
An array telescope that combines complete
sampling of the time, frequency and spatial domains
with a ×20-100 increase in collecting area (~ 1 km2)
over existing telescopes.
Frequency range 0.1 – 25 GHz (nominal)
Limited gains from reducing receiver noise or
increasing bandwidth once the EVLA is finished
Innovative design needed to reduce cost
• 106 meter2 ⇒ ~ €1,000 per meter2
• c.f. existing arrays ~ €10,000 per meter2
•
•
An international project from the start
International funding
•
17-country international consortium
•
Timeline for construction extends to 2020
• Can be phased for different frequency ranges
• Can do science as you go
•
The existing US radio astronomy portfolio is the
foundation on which to build the SKA
Number
Comments
Parallaxes:
Interferometry
timing
optical
×20
×50
Associations
Cost goal ~ € 1 billion
• Executive, engineering, science, siting, simulation
groups
•
Type
HI absorption
DM + ne model
1 mas @ 1 kpc
1.6 µs @ 1 kpc
HST, point spread function
~13
~5
~1
SNRs
8
GCs
16
LMC,SMC ~8
74
false associations
bright pulsars, galactic
rotation model
all radio pulsars
(~ 1400)
ISM perturbations
NE2001: Galactic Distribution of Free
Electrons + Fluctuations
Paper I = the model (astro-ph/0207156)
Paper II = methodology &
particular lines of sight (astroph/0301598)
Based on ~ 1500 lines of sight to
pulsars and extragalactic objects
Code + driver files + papers:
www.astro.cornell.edu/~cordes/NE2001
14
But … if you want a good distance,
measure the parallax !
Very Long
Baseline Array
PSR B0919+06
S. Chatterjee et al. (2001)
µ = 88.5 ± 0.13 mas/yr
π = 0.83 ± 0.13 mas
D = 1.2kpc
V = 505 km/s
e.g. Arecibo + GBT + VLAϕ
+
VLBA
will be a powerful parallax and
proper motion machine
Proper motion
and parallax
using the VLBA
(Brisken et al.
2001)
PSR B1929+10
Chatterjee
et al. 2003
15